Barrier Methods for Nonlinear Programming,
Abstract
A concise introduction and comprehensive survey is given in parametric auxiliary function methods for nonlinear programming, with emphasis on barrier function methods. An intuitive introduction is provided, general convergence results are proved, and important connections between these and well-known first and second order optimality conditions are detailed. Perturbation analysis results are summarized that provide a rigorous basis for convergence acceleration techniques. Algorithmic considerations are discussed in detail. Based on an exhaustive literature search, most of the significant variants and extensions of the basic approach that have been discovered to date are described, including various composite algorithms involving barrier and penalty functions, the generalized Lagrangian interpretation, exact penalty functions, differential gradient approaches, and hybrid algorithms designed to accelerate convergence and counter the ill-conditioning problem. Computational results and practical applications are indicated and potentially fruitful topics for future research are discussed. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 08, 1973
- Accession Number
- AD0772121
Entities
People
- Anthony V. Fiacco
Organizations
- George Washington University